The capability for automatic identification and segmentation of a user's hand from an image is useful in a variety of computer vision applications. Existing hand segmentation systems typically perform this process based on an assumption that the palm of the hand can be identified as a globally largest circular distention in an image that includes the user's hand and forearm, and thus the palm region can be approximated with a bounded circle or ellipse. This technique is not robust, however, as different hand shapes and orientations can result in the selection of a largest circular region in some other portion of the user's forearm, causing incorrect hand segmentation which can cause downstream computer vision tasks to fail.
Although the following Detailed Description will proceed with reference being made to illustrative embodiments, many alternatives, modifications, and variations thereof will be apparent in light of this disclosure.